Remarkable Performance Improvement (F17 Alpha)

[Palooka]

Remarkable performance improvement (x86_64) today. I assume that this is due to rc6 of the 3.3 kernel having dropped a ton of debugging code.

Great job, devs. Very robust too, apart from a couple of dependency issues, which are worked around quite easily,

Well done, guys!

[DBelton]

wish I could say that here! :lol:

After the update to the rc6 kernel, I get tons of messages thrown to my log file. They did fix an error but now I get tons of other messages. 3 message per second for however long the system is running.

Over and over, I am seeing this in the log file :(


Mar 7 15:34:35 tower20 kernel: [ 2411.517041] Reenabling IRQ 23
Mar 7 15:34:36 tower20 kernel: [ 2412.611370] Disabling IRQ 23
Mar 7 15:34:36 tower20 kernel: [ 2412.621038] Polling IRQ 23
Mar 7 15:34:37 tower20 kernel: [ 2413.621041] Reenabling IRQ 23
Mar 7 15:34:37 tower20 kernel: [ 2413.774292] Disabling IRQ 23
Mar 7 15:34:37 tower20 kernel: [ 2413.784032] Polling IRQ 23
Mar 7 15:34:38 tower20 kernel: [ 2414.784014] Reenabling IRQ 23
Mar 7 15:34:38 tower20 kernel: [ 2414.784146] Disabling IRQ 23
Mar 7 15:34:38 tower20 kernel: [ 2414.794019] Polling IRQ 23
Mar 7 15:34:39 tower20 kernel: [ 2415.794041] Reenabling IRQ 23
Mar 7 15:34:39 tower20 kernel: [ 2415.806208] Disabling IRQ 23
Mar 7 15:34:39 tower20 kernel: [ 2415.816037] Polling IRQ 23
Mar 7 15:34:40 tower20 kernel: [ 2416.816041] Reenabling IRQ 23
Mar 7 15:34:41 tower20 kernel: [ 2417.507136] Disabling IRQ 23
Mar 7 15:34:41 tower20 kernel: [ 2417.517036] Polling IRQ 23


I think this is the offending item, but not certain how to fix it



Mar 7 14:18:56 tower20 kernel: [ 0.648537] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
Mar 7 14:18:56 tower20 kernel: [ 0.648636] ehci_hcd 0000:00:1d.7: EHCI Host Controller
Mar 7 14:18:56 tower20 kernel: [ 0.648745] ehci_hcd 0000:00:1d.7: new USB bus registered, assigned bus number 1
Mar 7 14:18:56 tower20 kernel: [ 0.648853] ehci_hcd 0000:00:1d.7: using broken periodic workaround
Mar 7 14:18:56 tower20 kernel: [ 0.648920] ehci_hcd 0000:00:1d.7: debug port 1
Mar 7 14:18:56 tower20 kernel: [ 0.652861] ehci_hcd 0000:00:1d.7: irq 23, io mem 0xfea77c00
Mar 7 14:18:56 tower20 kernel: [ 0.658017] ehci_hcd 0000:00:1d.7: USB 2.0 started, EHCI 1.00
Mar 7 14:18:56 tower20 kernel: [ 0.658106] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
Mar 7 14:18:56 tower20 kernel: [ 0.658183] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
Mar 7 14:18:56 tower20 kernel: [ 0.658275] usb usb1: Product: EHCI Host Controller
Mar 7 14:18:56 tower20 kernel: [ 0.658335] usb usb1: Manufacturer: Linux 3.3.0-0.rc6.git0.2.fc17.x86_64 ehci_hcd
Mar 7 14:18:56 tower20 kernel: [ 0.658428] usb usb1: SerialNumber: 0000:00:1d.7

[Apostate]

I'm now using kernel-3.3.0-0.rc6.git0.2.fc17.x86_64 and my CPU usage has now dropped and stabilised - it was all over the place in previous f17 kernels.

[DBelton]

the rc6 kernel might help my cpu usage if it wasn't constantly writing to the log file due to the errors above :(

[Sylslay]

CONFIRM kernel-3.3. HAS BETTER PERFORMANCE IN OLDER FEDORA TOO.

[phoenixpb]

[phoenix@phoenix ~]$ uname -a
Linux phoenix.styx 3.3.1-3.fc17.x86_64 #1 SMP Wed Apr 4 18:13:49 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux

I confirm F17 is faster than F16 here. Great job.